[FLASH-USERS] Multigrid Poisson solver and different BC types on different boundaries

From: Artur Gawryszczak <gawrysz@camk.edu.pl>
Date: Fri Sep 05 2003 - 06:09:46 CDT

Hello,

I'm preparing a setup for self-gravitating disk in 3D (Cartesian grid). It
would be nice for such setup to have Neumann boundary conditions at a disk
plane and isolated b.c. elsewhere - this would speed up computations by a
factor at least 2. Can anybody tell me how to do it in best way, or warn me
about possible traps?

Here comes relevant part of my Config file:

REQUIRES mesh
REQUIRES mesh/amr/paramesh2.0/quadratic_cartesian
REQUIRES gravity/poisson/multigrid
REQUIRES solvers/poisson/multigrid
REQUIRES solvers/poisson/multigrid/fftcg

Such a combination seems to work, but I'm not sure if it's optimal.

I looked into Poisson solver related files, and I think, I should change
following ones:
        mg_init_src.F90 - add an exclusion in similar way as it's done for
'quadrant' parameter into a branch for MG_BND_GIVENVAL
        mg_set_ext_bndry.F90 - just change calculations made for 5th direction
        pois_solve_3d.f90 - force using real FFT and sacrifice performance.

Am I right, or did I miss something?

I'm not familiar with Poisson solvers, I'm just assuming that if single block
can handle different boundaries on its faces (mesh boundaries and neighbors)
it should be possible to have different gravitational b.c. even on
neighboring cells.

-- 
Thanks in advance,
        Artur
Received on Fri Sep 5 06:10:24 2003

This archive was generated by hypermail 2.1.8 : Thu Aug 31 2006 - 21:20:48 CDT